BURNOUT NUCLEATE CONVECTION WATER

advertisement
BURNOUT IN FORCED CONVECTION
NUCLEATE 80N
F WATER
BY J. M.. REYNOLDS
TECHNICAL REPORT NO. 10
OFFICE OF NAVAL RESEARCH
CONRACT N~ori-07894
NONR-1 848(39)
D.S.R. PROJECT NUMBER 7-7673
JY 1,1957
HEAT TRANSFER LABORATORY
MCANICAL ENGINEERING DEPARTMENT
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
CAMBRIDGE, MASSACHUSETTS
Summary
Data are presented for burnout in forced coivection nucleate boiling of
water at pressures above 500 psia. A dimensionless correlation is devised for.
the M.I.T. data which is found to be valid for certain recent data reported by
Argonne National Laboratory.
INTRODUCTION
The nucleate boiling process is characterized by an unstable maximum
heat flux which, if exceeded, results in a large discontinuous increase in
surface temperature.
The maximum or "burnout" heat flux is naturally of great
importance to the designer of apparatus utilizing a b&iling liquid as the
coolant. Recent experimental investigations of the phenomenon have yielded
sufficiently reproducible data to permit correlation of data from different
sources in certain flow regimes and system geometries. Burnout data recently
taken at M.I.T. is presented here along with a dimensionless correlation of
the forced convection burnout heat flux for water at pressures above 500 psia.
Description of the test aparatus
The test section used in this program consists of a 9-inch length of "A"
nickel tubing, .1805 inches in internal diameter and .2105 inches in external
diameter.
The tube is vertical and the flow upward. Electrical heating of
the test section is accomplished by passing current through the tube wall.
The source of electrical power is two 3000 amp 12 volt DC generators in series
which can be effectively controlled from zero to maximum output.
The system consists of a closed loop with a stainless steel DeLaval Hay~.ward-Tyler circulating pump. Pressure is maintained on the system by a
pressure veksel in which water is evaportted by electrical heating. The
pressure vessel is also used to degass the water before each run.
Distilled water is used in all runs and an ion exchanger is used to
maintain the impurity level below 1 ppm NaCl.
The test section instrumentation consists of inlet and outlet pressure
and voltage taps and nine chromel-constantan thermocouples on the tube outer
wall.
The wall thermocouples are insulated from the tube by .0015 in. mica.
The temperature difference across the tube wall is computed by the method of
ref 1. Inlet and outlet fluid temperatures are measured 5-1/2 inches upstream
end 9 inches downstream of the test section.
-2-
The tube is surrounded by four thermal shields, each enclosing the tube
for one-quarter of its length. The shields are individually variable resistance heaters wound on aluminum tubing, which are adjusted! to the average
temperature of the portion of the test section under each.
The requirements of the burnout program necessitate the use of a device
to prevent failure of the test section at the burnout point. This device, or
burnout detector, must be one which can identify the burnout condition and
take action to prevent failure, while allowing the experimentor to actually
In principle, the detector relies on the fact that
thermodynamic burnout occurs at the test section exit and is characterized
by a large temperature rise. Actually, the detector compares the electrical
attain burnout heat flux.
resistance of the final 1/16" of the test section with that of the remaining
These two resistances constitute two legs of a Wheatstone bridge
circuit, the other two legs consisting of fixed and variable resistors of
tube length.
At burnout the resistance of the final length of the test
section increases rapidly in relation to the remainder, causing an unbalance
in the bridge circuit.
suitable ratings.
The net voltage at the reference point in the bridge circuit provides
the input signal to a four stage DC amplifier.
The output of the amplifier
fires a triggering circuit which discharges a 7000 volt .10,uAf capacitor
through a spark gap in the circuit breaker.
The time required by the burnout detector to "decide" when a bona fide
burnout has occurred is relatively long, about 6 millisec.
At peak pressure
and heat flux, this interval is virtually all the time available before the
test section allowable overtemperature of 500 F is reached. Consequently,
an instantaneous circuit breaker is required to act on the signal from the
amplifier.
Rapid circuit interruption is accomplished by using the . 1 0 1"f capacitor
*
discharge to fire an explosive cap
which is contained in a hollow copper tube.
The test section current flows continually through the copper tube which is
one-half inch in outer diameter, 3-1/32" long, and has an .047" wall.
The
copper tube is restrained by a mechanical device which insures clean circuit
interruption when the cap is fired and low voltage drop through the tube and
*
DuPont X-98-N blasting cap, #6 strength, RDX loaded for 3000 F temperature
stability.
-3-
its connections during the run.
Details of the breaker construction as well
as a complete discussion of the detector circuitry will be subject of a project
report to be issued shortly.
The burnout detector has been an effective instrument for the study of
the burnout phenomenon and is well suited to the characteristics of the
process. During a run, the test section current is increased quite slowly.
Therefore, any bridge unbalance due to varying tube temperature can be readily
compensated by manual adjustment of the variable resistors. At high heat flux,
in the neighborhood of burnout, the test section is normally almost isothermal,
so that, while the magnitude of the test section resistance may vary, the resistances of the two portions of the tube remain in a relatively constant ratio
to each other. This males manual ad.4pstment unnecessary near the burnout point.
Almost without exception, the burnout detector has successfully prevented
failure of the test section and permitted the actual attainment of the burnout
condition in the test section. The fact that the detector is not firing prematurely is periodically checked by allowing a test section to fail while observing the time between the discharge of the capacitor and the appearance of
steam or other evidence of failure. These two events occur simultaneously.
TEST RESULTS
Burnout data were taken at 500, 1000, and 1500 psia system pressure and
at water inlet velocities of 5 and 10 feet per second. These data are presented in Appendix A.
The apparatus used in this work has been employed for a number of years
in studying the nucleate boiling process, and its limitations have been discussed in references 1 and 2. The following errors are inherent in the
measurement and computation schemes:
a. Heat flux is estimated to + 3%
b. Flow rate is estimated to + 2%
c. Inlet and outlet fluid temperatures and test
section wall temperatures are estimated to + 20 F.
These quantities are interrelated in three measures of heat rejection.
Heat
rejection may be determined from;
a. The difference in inlet and outlet fluid enthalpy as
measured by the corresponding temperatures and flow rate.
b. The product of current and test section voltage drop.
c. Summation of heat release in the nine wall sections
based on the current and tube resistivity at each of the nine
wall temperatures.
These three computations yield the + 3% heat flux variation.
Burnout data is recorded on a'l6-point Brown recorder since it is impossible to achieve steady state at the burnout point. A complete set of
data extends 16 seconds before burnout occurs so that some temperature readings must be estimated to + 30 F.
Current and voltage readings are printed
three times during each recorder cycle, so that errors in those readings are
not increased.
COhiELATION OF THE DATA
The Correlation
Correlation of the test datta is an obviously desirable goal which, once
attained, allows comparison of data from different sources. Unfortunately, the
forced convection burnout process is not yet sufficiently well understood to
allow one to use experimental and dimensional analysis techniques to full advantage.in determining correlation groups.
Nevertheless, it is hoped that
some progress may be made toward a correlation by the "brute force" technique
of fitting existing data with whatever combination of properties and system
parameters reduce scatter to an acceptable level. Later, these groups may be
investigated in the light of present knowledge and further experiments specifically designed to tdst the correlation.
In attempting to correlate the forced convection nucleate boiling process,
one must be somewhat skeptical regarding the usefulness of the normally defined heat transfer coefficient. Those familiar with the forced convection
boiling process will recognize that the heat transfer coefficient has meaning
only at a point and, when applied to a finite length of heat transfer surface,
becomes strongly dependent on the length uner consideration. In addition, a
-5useful film coefficient must not strongly depend on heat flux, as it does in
nucleate boiling.
A further difficulty lies in the measurement of T - Tb at high
pressure and high flux. Under these conditions, the temperature change across the
test section wall may be 3 or 4 times the wall-to-fluid temperature difference.
Such conditions enormously increase the difficulty of obtaining reliable data.
For these reasons, the present correlation was undertaken with the surface-toliquid heat transfer coefficient and temperature difference specifically excluded
from the list of possibly relevant parameters. In general, the properties to be
used for the correlation are restricted to those which do not involve especially
sentitive experimental techniques.
The problem of correlating the data bec6mes somewhat more manageable if
attention is directed to the region of burnout with net exit bulk quality. Under
these conditions, large variations in exit fluid properties abe obtained providing
an ideal area for testing groups based on properties at the burnout point. At the
same time, it appears reasonable to expect that estimates of flow conditions based
on average properties are more accurate in the quality region where large temperature gradients do not exist and vapor generation at the surface is at least
partially accounted for in the net steam generation predicted by average flow quality. For example, in the region of burnout with subcooled exit conditions it is
possible for vapor generation at the surface to significantly increase the exit
flow velocity over the value average exit fluid properties. Consequent,y, the
primary correlating effort was directed toward burnout with the average exit fluid
condition in the two-phase region. The following discussion applies exclusively
to the two phase flow regime.
If one computes the various exit flow paxameters, it becomes apparent that
the average exit velocity, assuming no slip between the phases, exhibits strong
potentialities as a correlating factor. Quite simple relations between burnhut
heat flux and exit velocity can be devised to correlate data over a reasonably
wide range of conditions. For example, a plot of vexit vs. q"/Klg correlates
the M.I.T. data reported in Appendix A as well as the more complicated final
result presented here. It is not possible, however, to extend such a simple
correlation outside the range covered by this particular set of data.
A second correlating factor may be found by reasoning that, in the absence
of large changes in slip ratio, the quantity of liquid adhering to the wall at
-6-
the exit section must become small at high quality,
Since the boiling
process depends on a liquid covered surface, it is reasonable to expect
that the average depth of the liquid film on the surface is an important
variable when this depth is small. Under these conditions, the maximum
bubble diameter would be limited by the liquid film thickness. Computation
of the film thickness based on all the liquid in the stream at average exit
quality yields encouraging results. This term,
, proves to be a strong
*
correlating factor.
Having V0 and $ available as correlating factors, dimensional reasoning leads one to a relation of the following type:
IF/D
_L
Correlation of the data with these groups results in the following specific
relation,
isq4
.
/0I-2
-)
(2)
Discussion of the Correlation
The results of applying the correlation to the M.I.T. and ANL data are
shown in Figures 1 and 2. The range of operating points covered by the data
correlated is:
a) Heat flux from 0.6 x 106 to 2.8 x 106 Btu/hr ft2
b) Flow rate from 0.66 x 106 to 3.6 x 106 lb/hr ft2
c) Pressure from 500 to 2000 psia
d) Average exit quality from 0.01 to 0.60
e) Test section L/D 50 and 76 (D of .18" and .306")
The correlation may be conveniently broken down to show the predicted
dependence of q" on the various flow parameters and water properties.
X /)
where
f(
)
where
=
(
(3)
(4)
-7From Fig. 3, one can see that the variation of f(x,p) is such that the correlation predicts a continuously decreasing peak flux with increasing pressure
(except very near the critical pressure). It is improbable that such a relationship is valid at low pressures. The limit of validity of this particular
group of properties has not been fully established; it appears to fit the experimental facts for pressures as low as 500 psia at flow rates as low as
1 x 106 lb/hr ft2
The correlation given by equation (3) indicates an inverse relationship
between burnout heat flux and flow rate at constant exit fluid properties.
This is contrary to the relationship reported in other operating regimes (ref.
3), specifically, with subcooled exit conditions. Both the direct and inverse
variation of q" with G appear well established in specific operating ranges,
but the dividing line between the two variations has not yet been clearly defined. The present correlation indicates that the inverse relationship is
valid over at least the following very roughly defined region in water systems.
a) Average exit bulk quality below 50%
b) Pressure above 500 psia
c) In the region of inlet subcooling below 2500 F on the M.I.T. test
section
The effect of approaching these limits may be seen in Fig. 2 in which a systematic deviation from the correlation line is observed at qualities greater
than 60% or !
.06.
-8Limitations of the correlation
The correlation presented in this paper is inherently restricted to the
prtdkction of burnout in water systems with vertical test sections and upflow.
In addition, it is specifically directed toward burnouts with net exit bulk
qualities. The correlation is presently untried for test section hydraulic
diameters outside the range .18" to .306", the L/D range from 50 to 76, and
pressures below 500 psia. In addition, the data correlated were taken with
DC electrical heating and burnout was induced by increasing heat flux at constant pressure, flow rate, and inlet temperature. To date, no data have been
taken to specifically test the correlation so that, in its present state, it
must be recognized as tentative.
In addition to the above limitations of the data correlated, the form of
the parameters involved have the characteristic of greatly amplifying experimental errors in certain ranges. These parameters are, to a certain extent,
self compensating. For example, a positive error in exit quality produces a
negative error in the ordinate and a negative error in the abscissa tending to
bring the point back on the curve. In general, increasing over-compensation
exists at high quality () 60%) and low pressure. Experimental errors in the
determination of exit quality are reflected in the correlating parameters
according to the following relations
vo )
-
4o
*
(5)
_
(6)
-
Vo
Errors in the direct measurements of heat flux and flow rate are reflected in
quality as a function of inlet subcooling.
dv
(7)
p/-
Where
AH
=
1/
-
is the enthalpy rise in the test section and
t b
is
the difference between saturated liquid enthalpy (hf) and fluid inlet enthalpy.
Prediction of peak heat flux from the correlation
Unfortunately, heat flux is implicit in
and "'P, so that it cannot be
read directly from the curve for a given pressure and flow rate. A trial and
error computation is required. However, experience with the correlation indicates that three trials are sufficient for an accurate graphical solution. Convergence on the predicted value is rapid and the value itself clearly defined.
References
1. Rohsenow, W. M., Somma, E. H., and Osborn, P. V., "Construction and
Operation of Apparatus for study of Heat Transfer with Surface Boiling",
Technical Report No. 2, DIC Project No. 6627, Mass. Institute of Technology, July 1, 1950.
2. Clark, J. A., and Rohsenow, W. M., "Local Boiling Heat Transfer to Water
at Low Reynolds Numbers and High Pressures", Transactions of the ASME'
Vol. 76, No. 4, May 1954.
3.
Jens, W. H., and Lottes, P. A., "Analysis of Heat Transfer, Burnout,
Pressure Drop and Density Data for High-Pressure Water",
Argonne National
Laboratory, ANL-4627, May 1, 1951.
4.
Weatherhead,
R. J., "Burnout with Net Steam Generation in Vertical, Round
and Rectangular Channels," Argonne National Laboratory.
Presented at the
Reactor Heat Transfer Conference, New Yos New York, November 1, 1956.
Nomenclature
q"
G
heat flux, ft-lbf/ft -hr
specific flow rate, lbm/hr-ft 2
D
inner diameter of test section, ft
R
D/2
V0
test section exit velocity based on average exit fluid condition,
ft/hr
test section heated length, ft
change in enthalpy of a unit mass flowing through the test section,
ft-lbf/lbm
saturated liquid enthalpy minus test section inlet enthalpy,
ft-lbf/lbm
heat of vaporization ft-lbf/lbm
L
A H
A h
hfg
g
g
conversion factor, ft-lbm/lbf-sec 2
gravitational acceleration, ft/hr2
average outlet density, lbm/ft3
saturated liquid density, lbm/ft 3
(T*
surface tension at saturation temperature, lbf/ft
/4,
saturated liquid 'viscosity, lbm/ft-hr
x
mass quality, dimensionless
y
v
vf
volume quality = xvy/(vf + xvfg), dimensionless
saturated vapor specific volume, ft 3 /lbm
saturated liquid specific volume, ft 3 /lbm
fg
g
f
height of liquid film on heating surface at average exit fluid
condition with no slip between phases, ft
APPENDIX A
MIT Burnout Data
Inlet
Temp (P)
Pressure
.(P sia) -
264
249
298
308
540
372
516
400
414
395
380
461
462
249
291
314
356
387
427
394
410
465
462
232
253
279
312
347
409
421
424
542
509
463
476
474
485
259
357
318
545
555
545
565
575
555
585
585
555
535
535
545
565
530
575
545
535
539
527
1035
1010
981
1020
1025
1065
1053
1015
.1020
1020
1025
1005
1005
1035
999
1015
1065
985
343
323
336
436
1015
1040
499
497
420
461
1010
1010
1055
1015
993
Flow Rate
6
(b/rft2) 10" _
Heat Flux .6
(Btu/hr-ft ) 10"
Exit
uality
2.10
2.12
2.86
2.80
.036
2.13
2.59
2*54
.057
2.04
1.97
2.01
1.93
2.00
1.94
1.86
1.86
1.04
1.02
1.02
1.10
0,962
0.971
0.990
0.936
0.936
0.939
2.13
2.13
2.08
2.05
2.02
1.88
1.91
1.91
1.70
1.80
1.85
1.77
1.64
1.84
1-04
1.00
1.02
0.965
1.02
1.01
0.961
0.886
0.889
0.964
0.950
2.22
2.14
2.12
2*11
2.07
1.90
1.87
2.30
2*19
2.09
2.05
2.01
2.00
1.98
1.92
1.82
1.82
2*81
.029
.089
.153
.157
.180
.173
.139
.237
.230
.260
.297
.312
.325
.393
.454
.418
.422
.470
.466
Sub.
2*56
2.55
2.39
2*18
1.97
-072
1 .82
.053
1-80
1.67
1.62
1.60
1-53
1.53
1.47
2.08
1.84
1.83
1 82
1.81
1.79
1.62
1*52
1.51
1.51
1.49
.062
.196
.208
.123
.145
.155
*138
.142
.238
.160
.196
.183
.190
.291
.440
.428
.254
.354
Appendix A (continued)
Inlet
Te2,M (OF)
Pressure
(Psia)
Flow Rate
6
(lbM/r-ft2) 10
502
538
396
509
544
388
438
464
434
495
509
530
320
375
352
382
400
454
495
506
516
1009
985
975
1015
1045
1525
1525
1525
1533
1525
1515
1520
1510
1505
1485
1505
1510
1515
1510
1550
1511
0.886
0.866
0.895
0.890
0.859
1.94
1.89
1.88
1.89
1.79
1.76
1.76
1.02
0.990
0.986
0.964
0.954
0.938
0.881
0.882
0.860
Heat gu06
(Bu/rft4)
1.43
1.41
1.39
1.32
1.29
1.97
1089
1.60
10
Exit
Quaity
.406
.464
*381
.409
.443
Subc.
It
.006
1059
subc.
1.45
1.37
1.25
1.67
1.61
1.52
1.47
1.45
1033
1.21
1.15
1.14
.056
.076
.112
.019
.122
.059
.097
.127
.196
.270
. 241
.292
APPENDIX B
Inlet
Temp (OF)
625
613
636
595
633
586
590
627
614
635
608
613
619
587
622
623
616
613
611
612
597
598
581
624
610
624
ARGONNE NATIONAL LABORATORY
Burnout Data, .18" Tube
2000 psia, L/D = 64.5
(Ref. 4)
Flow Rate
Heat Flux
(lb/hr-fta) 10~0
(Btu/hr-ft2) 103.50
3.56
3.44
2.57
2.94
3.30
3.68
3.50
3.58
3.47
2.54
2.51
3.00
3.58
2.71
2.68
2.74
2.88
2.94
2.90
3.05
3.10
3.22
2.04
1.89
1.86
0.850
1.00
0.745
Exit
Quality
.094
.074
.121
0.995
.077
0.682
.118
1.15
.028
.016
.095
.077
.115
1.15
0.820
1.00
0.742
0.863
0.845
0.820
1.14
0.746
0.735
0.805
0.799
0.850
0.840
0.990
1.00
1.15
0.658
0.660
0.660
.091
.104
.090
.014
.101
.104
.091
.071
.072
.075
.048
.049
.019
.136
.102
.151
Appendix B (continued)
ARGONNE NATIONAL LABORATORY
Burnout Data .306" Tube
2000 psia, L/D = 76
(Ref. 4)
0.186
0.188
0.167
0.210
0.241
0.183
0.304
0.314
0.353
0.230
0.246
0.435
0.268
0.226
0.241
0.398
0.505
0.61W
0.725
0.441
0.176
.495
631
609
628
633
614
632
584
629
525
525
633
510
504
513
599
618
507
523
580
0.169
0.198
0.196
0.215
0.258
0.245
0.275
0.278
0.328
0.327
0.318
0.360
0.357
0.356
0.356
0.346
0.587
0.600
0.393
.574
.676
.580
.580
.840
.515
.407
.491
.619
.560
.470
.525
.670
.620
.459
.384
.262
.218
.403
465
627
0.534
0.550
0.595
0.346
.380
600
23
10
4
*
o
X
2 -
2000 psi, Argonne Not. Lab. Data
1500 psi
M. I.T. Data
1000 psi
500psi
/
--
/
J
2
2
10
6
2
4
e_/#A/
__.
0
+ 30 %Z-
,0'
30 % Z
I
C) 0
b
-x
2
-1/-
4
/4/
/s
s.
4
IO
-9 =|49 x10 Z
/ /
6
2
/9
x
I
.01
1.
4
2
6
4
.10
6
!.0
8 /R
FIG. I-
CORRELATION
IN .18"
OF BURNOUT
1. D. TUBES.
HEAT FLUX
FOR WATER
22
10
0
o
C
\0
21
-o- b
N
.01
2
4
6
8
.10
2
4
6
8
8/R3
FIG. 2-
CORRELATION
I.D. TUBE.
OF BURNOUT
ARGONNE
HEAT
NATIONAL
FLUX
WATER
LABORATORY
IN .306"
DATA
1.0
-1I
10
2
b
>,
x =0.25
10
8
4
0.5
42
1.49xO
q"G
1.5
D
f(x, p)g,
2
-13
10
FIG. 3-
500
0
DEPENDENCE
f (x,p)
vs.
1000
Pressure (psia)
OF CORRELATION
PRESSURE
1500
ON WATER
2000
PROPERTIES
Download